In the past, many companies were shocked when they looked at the billing after the initial enthusiasm about how convenient it is to rent computing power from cloud providers. When rental resource costs skyrocket, the search for culprits begins. But which pods in a Kubernetes cluster are really driving up costs by consuming RAM, CPU, disk space, and network traffic isn’t obvious at first glance. The open source project OpenCost started in mid-2022 to provide such insights. To the Microsoft has now also announced KubeCon 2023to actively participate and immediately contributed functions.
OpenCost was originally invented by the company Kubecost, which continues to offer a hosted version. Under the name OpenCost, the software became an open source project of the Cloud Native Computing Foundation (CNCF) and is in sandbox status there. Among the founding members were two of the largest cloud providers, Amazon AWS and Google Cloud. Almost half a year later, Microsoft completes the round.
OpenCost works in conjunction with the metrics collector Prometheus, which can collect detailed information on the resources each individual pod is using. OpenCost combines these observations with a machine-readable price list for the cloud provider and generates accurate reports.
New function
Microsoft is bringing two new functions to the debut. On the one hand, the company has integrated a connection to its own Azure price API. In future, access data for an Azure account will be stored in the OpenCost instance so that the software can query the exact costs for the resources used. This makes the information more accurate, because individual price lists that large customers have negotiated with Microsoft sales are also taken into account. In OpenCost, this feature ends up in version 1.103, which is scheduled to be released on May 2nd, 2023. If you want to test them by then, you have to Install KubeCon pre-release.
The second function is only indirectly related to Azure. Microsoft has built a CSV exporter that compiles a daily evaluation of the costs incurred in CSV format and stores it locally or in Azure storage, S3 or Google storage. Such exports are intended, for example, for non-technical departments in the company.
(jam)
